Outline




Munge

Load packages
using pacman

library(pacman)
p_load(knitr,  # for weaving this into pretty format
       readr,  # for read_csv()
       tibble,  # for an easier way to work with data.frames
       dplyr,  # for data manipulation
       stringr,  # for string manipulation
       lubridate,  # for dates
       tidyr,  # for gather() and spread()
       data.table,  # for 
       DT,  # for kable()
       ggplot2,  # for plots
       ggrepel  # for spreading point labels
       )

Import data
using readr

bring_heat_2 <- read_csv('./ChicagoWildfire2015-stats.csv')

Make the data into a tibble

bh <- as_tibble(bring_heat_2)

Take a look at what we have

bh
## # A tibble: 5,132 × 42
##            `Date/Time` Tournamemnt        Opponent `Point Elapsed Seconds`
##                 <dttm>       <chr>           <chr>                   <int>
## 1  2015-06-12 20:31:00        <NA> vs Thunderbirds                     133
## 2  2015-06-12 20:31:00        <NA> vs Thunderbirds                     133
## 3  2015-06-12 20:31:00        <NA> vs Thunderbirds                     133
## 4  2015-06-12 20:31:00        <NA> vs Thunderbirds                     133
## 5  2015-06-12 20:31:00        <NA> vs Thunderbirds                     133
## 6  2015-06-12 20:31:00        <NA> vs Thunderbirds                     133
## 7  2015-06-12 20:31:00        <NA> vs Thunderbirds                     133
## 8  2015-06-12 20:31:00        <NA> vs Thunderbirds                     133
## 9  2015-06-12 20:31:00        <NA> vs Thunderbirds                     133
## 10 2015-06-12 20:31:00        <NA> vs Thunderbirds                     133
## # ... with 5,122 more rows, and 38 more variables: Line <chr>, `Our Score
## #   - End of Point` <int>, `Their Score - End of Point` <int>, `Event
## #   Type` <chr>, Action <chr>, Passer <chr>, Receiver <chr>,
## #   Defender <chr>, `Hang Time (secs)` <dbl>, `Player 0` <chr>, `Player
## #   1` <chr>, `Player 2` <chr>, `Player 3` <chr>, `Player 4` <chr>,
## #   `Player 5` <chr>, `Player 6` <chr>, `Player 7` <chr>, `Player
## #   8` <chr>, `Player 9` <chr>, `Player 10` <chr>, `Player 11` <chr>,
## #   `Player 12` <chr>, `Player 13` <chr>, `Player 14` <chr>, `Player
## #   15` <chr>, `Player 16` <chr>, `Player 17` <chr>, `Player 18` <chr>,
## #   `Player 19` <chr>, `Player 20` <chr>, `Player 21` <chr>, `Player
## #   22` <chr>, `Player 23` <chr>, `Player 24` <chr>, `Player 25` <chr>,
## #   `Player 26` <chr>, `Player 27` <chr>, `Elapsed Time (secs)` <int>

^ yikes




Manipulate variable names
Check out column names

names(bh)
##  [1] "Date/Time"                  "Tournamemnt"               
##  [3] "Opponent"                   "Point Elapsed Seconds"     
##  [5] "Line"                       "Our Score - End of Point"  
##  [7] "Their Score - End of Point" "Event Type"                
##  [9] "Action"                     "Passer"                    
## [11] "Receiver"                   "Defender"                  
## [13] "Hang Time (secs)"           "Player 0"                  
## [15] "Player 1"                   "Player 2"                  
## [17] "Player 3"                   "Player 4"                  
## [19] "Player 5"                   "Player 6"                  
## [21] "Player 7"                   "Player 8"                  
## [23] "Player 9"                   "Player 10"                 
## [25] "Player 11"                  "Player 12"                 
## [27] "Player 13"                  "Player 14"                 
## [29] "Player 15"                  "Player 16"                 
## [31] "Player 17"                  "Player 18"                 
## [33] "Player 19"                  "Player 20"                 
## [35] "Player 21"                  "Player 22"                 
## [37] "Player 23"                  "Player 24"                 
## [39] "Player 25"                  "Player 26"                 
## [41] "Player 27"                  "Elapsed Time (secs)"

Get rid of players 8 through 27 because they will never exist.
(There are only ever 7 players per team on the field at once.)

bh <- bh %>%
  select(
    -(`Player 8`:`Player 27`)
  )

Replace all spaces in column names with underscores

names(bh) <- str_replace_all(names(bh), " ", "_")

Take out “- End of Point” from Their Score - End of Point and Our Score - End of Point

names(bh) <- str_replace_all(names(bh), '_-_End_of_Point', '')

Take out the seconds unit at the end of hang time and elapsed time

names(bh) <- gsub('_(secs)', '', names(bh), fixed = TRUE) # a bit easier with gsub than with str_replace because of the parentheses around secs

Make all variable names lowercase

names(bh) <- tolower(names(bh))

Rename date/time column to date_time

bh <- bh %>% 
  rename(
    date_time = `date/time`
  )

Now our column names are more manageable: 

names(bh)
##  [1] "date_time"             "tournamemnt"          
##  [3] "opponent"              "point_elapsed_seconds"
##  [5] "line"                  "our_score"            
##  [7] "their_score"           "event_type"           
##  [9] "action"                "passer"               
## [11] "receiver"              "defender"             
## [13] "hang_time"             "player_0"             
## [15] "player_1"              "player_2"             
## [17] "player_3"              "player_4"             
## [19] "player_5"              "player_6"             
## [21] "player_7"              "elapsed_time"




Manipulate variable types
Make our categorical variables into factors

# make vector of columns we want as factor
want.as.factor <- c('tournamemnt', 'opponent', 'line', 'event_type', 'action', 
                    'passer', 'receiver', 'defender',
                    'player_0', 'player_1', 'player_2', 'player_3', 'player_4', 'player_5',
                    'player_6', 'player_7')


# make these columns factors
bh[, want.as.factor] <- data.frame(apply(bh[, want.as.factor], 2, as.factor))

Make our numeric variables numeric

# make vector of columns we want as numeric
want.as.numeric <- c('point_elapsed_seconds', 'our_score', 'their_score', 
                     'hang_time', 'elapsed_time')

# make these columns numeric
bh[, want.as.numeric] <- data.frame(apply(bh[, want.as.numeric], 2, as.numeric))

Check that tournament is always the same and take it out

summary(bh$tournamemnt) # careful of the typo -- variable name is tournamenmt
## NA's 
## 5132
# it's always NA so take it out
bh <- bh %>%
  select(
    -tournamemnt
  )




Clean up the opponent column

Check the levels in the opponent column

levels(bh$opponent)
##  [1] "Alleycats"       "AlleyCats"       "Mechanix"       
##  [4] "Radicals"        "Revolution"      "Thunderbirds"   
##  [7] "vs Thunderbirds" "vs. Radicals"    "Vs. Radicals"   
## [10] "Windchill"

There are multiple columns for the same opponent

Take “vs.” in all its forms out of the opponents column

levels(bh$opponent) <- levels(bh$opponent) %>%
                                 str_replace_all(c('vs ', 'vs. ', 'Vs. '), '')

Check that our levels of opponent look right

levels(bh$opponent)
## [1] "Alleycats"    "AlleyCats"    "Mechanix"     "Radicals"    
## [5] "Revolution"   "Thunderbirds" "Windchill"

We have two types of Alleycats: Alleycats and AlleyCats. Combine these levels.

bh$opponent[bh$opponent=='AlleyCats'] <- 'Alleycats'

And drop the unused level

bh <- bh %>%
  droplevels()

Keep all columns in a bh_all tibble so we can safely pare down bh

bh_all <- bh




Tidy


Split date_time into date and time columns so that we can differentiate different games played against the same opponent.
(These will have the same opponent name an different dates.)

bh_all <- separate(bh_all, date_time, into = c("date", "time"), sep=" ")

Get date in date format

bh_all$date <- ymd(bh_all$date)

Check that class is “Date”

class(bh_all$date)
## [1] "Date"

The time column is the same for each date (gives us no more information than date), so we can safely ignore / get rid of it



Game numbers

Make a column for each game number against each opponent

bh_all <- bh_all %>%
  group_by(opponent) %>%
  mutate(
    game = as.numeric(factor(date))
  )

bh_all$game <- factor(bh_all$game) # make game a factor

Make a column for each individual game

library(data.table)
bh_all <- setDT(bh_all)[, ind.gme := .GRP, by = .(date, opponent), ]

Turn bh_all back into tibble

bh_all <- as_tibble(bh_all)

Check that our last ind.gme value is 13

tail(bh_all[, c(1, 3, 23:24)])
## # A tibble: 6 × 4
##         date opponent   game ind.gme
##       <date>   <fctr> <fctr>   <int>
## 1 2015-05-20 Radicals      1      13
## 2 2015-05-20 Radicals      1      13
## 3 2015-05-20 Radicals      1      13
## 4 2015-05-20 Radicals      1      13
## 5 2015-05-20 Radicals      1      13
## 6 2015-05-20 Radicals      1      13



Pare bh down to columns we really care about

bh <- bh_all

bh <- bh %>%
  select (
    opponent, our_score, their_score,
    action, 
    passer, receiver, defender, 
    elapsed_time, game, ind.gme
  )

Gather player into one column

Combine passer, reciever, and defender into a new column, actor.
Make a name column corresponding to name each row in actor.

bh <- bh %>% 
  tidyr::gather(
    key = actor,
    value = name,
    passer:defender,
    na.rm=T
  )

head(bh[, 4:9])
## # A tibble: 6 × 6
##   action elapsed_time   game ind.gme  actor   name
##   <fctr>        <dbl> <fctr>   <int>  <chr>  <chr>
## 1  Catch           34      1       1 passer Fergus
## 2  Catch           38      1       1 passer    Bob
## 3  Catch           41      1       1 passer Fergus
## 4  Catch           46      1       1 passer  Shiel
## 5  Catch           50      1       1 passer Cullen
## 6  Catch           54      1       1 passer Adrian

Make actor and name factors

bh$actor <- factor(bh$actor)
bh$name <- factor(bh$name)




Get summary stats

Calculate plus minus per name (i.e., player) as

\[ p\_m = (goals + blocks) - (throwaways + drops) \]

and arrange names by p_m in descending order.

plus_minus <- bh %>%
  select(
    name, action
  ) %>%
  group_by(
    name
  ) %>%
  summarise(
    goals = sum(action == 'Goal'),
    Ds = sum(action == 'D'),
    throwaways = sum(action == 'Throwaway'),
    drops = sum(action == 'Drop'),
    p_m = (goals + Ds) - (throwaways + drops)
  ) %>%
  arrange(desc(
    p_m
  ))

Take out the player “Anonymous” [he is probably a scary hacker]

mvplus_minus <- plus_minus %>% 
  filter(name != "Anonymous")

Print the table with players’ summary stats.
p_m is each player’s plus minus.




kable(mvplus_minus, format = "markdown")
name goals Ds throwaways drops p_m
AJ 42 8 7 3 40
Shiel 43 5 12 2 34
Goose 62 12 31 15 28
Kaufmann 23 8 5 4 22
Slymer 20 10 6 2 22
Danny 28 4 3 9 20
Uber 14 11 5 2 18
DW 12 7 1 3 15
Matzuka 45 18 31 17 15
Cullen 42 9 25 12 14
Vock 12 6 2 4 12
Fergus 33 6 20 8 11
Jack 15 7 5 6 11
Neil 13 5 4 3 11
JRob 7 8 7 0 8
Rathunde 7 5 3 1 8
Bob 44 8 34 11 7
MattSteg 7 2 2 0 7
Rory 10 0 2 1 7
Max 6 8 4 4 6
Sheehan 15 10 18 1 6
J Silver 4 3 2 0 5
Jimmy 3 3 1 0 5
Von 5 1 1 0 5
JSS 5 6 3 6 2
Roush 1 1 0 0 2
Silver 2 0 0 0 2
Brodie 1 0 1 0 0
Kennedy 13 4 14 3 0
Zubair 5 1 4 2 0
Adrian 23 4 15 13 -1




Plot


Approximate the Wildfire MVP by plotting season Ds vs. season goals.

mvp_plot_2 <- ggplot(data = mvplus_minus, aes(x = Ds, y = goals, 
                                   label = name, angle = 45))
mvp_plot_2 + geom_point() +
  ggtitle("Chicago Wildfire 2015 MVP") +
  xlab("Season Blocks") + 
  ylab("Season Goals") +
  geom_text_repel(aes(label = name), 
                  box.padding = unit(0.45, "lines"))


Disclaimer: this plot doesn’t consider the bad things [drops and throwaways], so the p_m column of the mvplus_minus table is probably a more accurate indicator of a player’s contributions.
It also does not correct for differences in playing time per game or number of games played.